This is a Google SoC project (see announcement).  It is currently at release 0.1:

Description

The software gnome-voice-control is used to control the GNOME desktop by voice, his first release implements action like run terminal, run mail, run browser,maximize window, minimize window, close window, next window.

Download

http://www.comp.ufscar.br/~raphael18/gnome-voice-control-0.1.tar.gz

It uses the Sphinx-2 speech recognition engine.

Ken 

--- (Edited on 6/20/2007 9:49 am [GMT-0400] by kmaclean) ---

For demo see

http://raphaelnunes.wordpress.com/2007/06/16/gnome-voice-control-demonstration/

Current SVN trunk

 svn checkout http://svn.berlios.de/svnroot/repos/festlang/trunk/gnome-voice-control

also utilizes GNOME ally framework and allows you to control UI with speech. For example if dialog appeared you can just say "Cancel" and dialog will be closed. Or you can say "File .. Open" to run "Open file" dialog. So you can just navigate menus. Release 0.2 with this functionality will be available very soon.

About sphinx2, we are planning to switch to sphinx3 actually to support French, Spanish and Russian. 

--- (Edited on 6/20/2007 9:20 am [GMT-0500] by nsh) ---

 I just rewatched both demonstration videos of Gnome-voice-control and I truly think that Raphael did excellent work.

 I also think that it would be very nice if we could find a native speaker to make a new screencast. perhaps someone on the Gnome-team or at Ubuntu? A native speaker would be better able to demonstrate the quality of this great piece of work.

 Ideally this screencast should end with a message like " help improve speech recognition on Linux: Donate some speech at VoxForge.org" or something like that.

If I knew how, I would definitely offer to make it myself. Though I think that I also do not qualify as a native speaker.

On that subject, I was also wondering if there was any progress on making gnome-voice-control suitable for other languages?

--- (Edited on 11/3/2007 6:09 am [GMT-0500] by Robin) ---

> I just rewatched both demonstration videos of Gnome-voice-control and I truly think that Raphael did excellent work.

Thanks

>  Ideally this screencast should end with a message like " help improve speech recognition on Linux: Donate some speech at VoxForge.org" or something like that.

Good idea really, we have to find someone. Ken will help probably :) Moreover with a recent release 0.3 we have something new to show.

> On that subject, I was also wondering if there was any progress on making gnome-voice-control suitable for other languages?

Well, technically there is no big problem, we only need a new model. French model is available for example and thus French support stays on the top on my TODO on gvc.

--- (Edited on 11/3/2007 8:35 pm [GMT-0500] by Visitor) ---

Hi NSH,

>>  Ideally this screencast should end with a message like " help improve

>>speech recognition on Linux: Donate some speech at VoxForge.org" or >>something like that.

>Good idea really, we have to find someone. Ken will help probably :)

>Moreover with a recent release 0.3 we have something new to show.

I can help ... though I would not say my voice is radio-announcer quality :) 

Is there specific software I should look at to create such a demo?

Ken

--- (Edited on 11/6/2007 12:03 am [GMT-0500] by kmaclean) ---

Hm, you need GNOME desktop, gtkrecordmydesktop and gvc of course. I'll try to create a scenario then. Will be ready soon.

--- (Edited on 11/6/2007 2:09 am [GMT-0600] by nsh) ---

gvc has just got German and Russian support. To test it you need to checkout gvc from svn trunk as well as pocketsphinx and sphinxbase from trunk. Then unpack voxforge acoustic model to ${prefix}/share/pocketsphinx/models/hmm/voxforge-de and start applet in de_DE locale. Everything should work fine, testers are welcome.

--- (Edited on 5/1/2008 11:35 am [GMT-0500] by nsh) ---